> whether laryngeal-induced coloring (and lengthening) is a comparativ or
internal reconstruction.
Many language groups merge the effects together, so that the most their
evidence can tell us is that there is a laryngeal of some sort present. The
prothetic vowel in Greek seems at times to reflect the nature of the root
vowel (compare edo: eat and odont- tooth), but more often Greek does seem
to reflect one of the three vowel-colouring laryngeals. Sometimes, though,
only Greek allows us to tell the difference, and that's not an ideal
situation. And although many people write of three laryngeals, there is
still no complete agreement. Some people want 4 (or more), others want
fewer.
.....Another note, is there any evidence outside of >Greek for root-initial
preconsonantal (ie. vocalizing) laryngeals?
Yes, Armenian (and sometimes Phrygian, it is claimed) also show the
"prothetic" vowel.
>The idea of various complex onset clusters reconstructed on the evidence of
one branch only sounds distressingly ad hoc to me.
Yes, it would be. And if that were the case, it would be better to
reconstruct something internal to Greek. But there is a variety of
fragments of evidence from other languages indicating the presence of an
initial laryngeal on many of these words. In the few cases when there is no
such evidence, or even evidence against it, some people suggest that Greek
has merely added an etymologically unjustified initial vowel, for one reason
or another. So the mere presence of a prothetic vowel in Greek is not
enough.
> is there any consonantal segmental evidence for root-initial prevocalic
*h1?
Yes. Vowel lengthening in a previous syllable is sometimes explained that
way. You find this in compound forms, and in verse. There is the
occasional suggestion of this in Homer, though it seems unlikely. The
examples from the Vedas, especially the Rg Veda, are more likely. Compounds
are better.
e.g. h1widheu "separate" Pokorny p1127 Skt vidhyati pierce has
reduplicated aor avi:vidhat with lengthened i:
and h1 ger "wake" which I mentioned in a previous post, also showing
lengthening in a reduplicated form in Skt.
Peter